检索结果-内蒙古大学图书馆

AsyCo: an asymmetric dual-task co-training model for partial-label learning

science China(Information sciences) 2025年第5期68卷 332-347页

作者： Beibei LI Yiyuan ZHENG Beihong JIN Tao XIANG Haobo WANG Lei FENG College of Computer Science Chongqing University State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences University of Chinese Academy of Sciences School of Software Technology Zhejiang University School of Computer Science and Engineering Nanyang Technological University

Partial-label learning(PLL) is a typical problem of weakly supervised learning, where each training instance is annotated with a set of candidate labels. Self-training PLL models achieve state-of-the-art performance but suffer from error accumulation problems caused by mistakenly disambiguated instances. Although co-training can alleviate this issue by training two networks simultaneously and allowing them to interact with each other, most existing co-training methods train two structurally identical networks with the same task, i.e., are symmetric, rendering it insufficient for them to correct each other due to their similar limitations. Therefore, in this paper, we propose an asymmetric dual-task co-training PLL model called AsyCo,which forces its two networks, i.e., a disambiguation network and an auxiliary network, to learn from different views explicitly by optimizing distinct tasks. Specifically, the disambiguation network is trained with a self-training PLL task to learn label confidence, while the auxiliary network is trained in a supervised learning paradigm to learn from the noisy pairwise similarity labels that are constructed according to the learned label confidence. Finally, the error accumulation problem is mitigated via information distillation and confidence refinement. Extensive experiments on both uniform and instance-dependent partially labeled datasets demonstrate the effectiveness of AsyCo.

关键词： machine learning weakly supervised learning partial-label learning co-training models candidate label sets

来源：评论

学校读者我要写书评

暂无评论

Deep learning-based software engineering: progress,challenges, and opportunities

引用

science China(Information sciences) 2025年第1期68卷 57-144页

作者： Xiangping CHEN Xing HU Yuan HUANG He JIANG Weixing JI Yanjie JIANG Yanyan JIANG Bo LIU Hui LIU Xiaochen LI Xiaoli LIAN Guozhu MENG Xin PENG Hailong SUN Lin SHI Bo WANG Chong WANG Jiayi WANG Tiantian WANG Jifeng XUAN Xin XIA Yibiao YANG Yixin YANG Li ZHANG Yuming ZHOU Lu ZHANG School of Journalism and Communication Sun Yat-sen University School of Software Technology Zhejiang University School of Software Engineering Sun Yat-sen University School of Software Dalian University of Technology School of Computer Science and Technology Beijing Institute of Technology Key Laboratory of High Confidence Software Technologies (Peking University) Ministry of EducationSchool of Computer Science Peking University State Key Laboratory for Novel Software Technology Nanjing University School of Computer Science and Engineering Beihang University Institute of Information Engineering Chinese Academy of Sciences School of Computer Science Fudan University State Key Laboratory of Complex & Critical Software Environment (CCSE) School of Software Beihang University School of Computer and Information Technology Beijing Jiaotong University School of Computer Science and Technology Harbin Institute of Technology School of Computer Science Wuhan University Huawei Technologies

Researchers have recently achieved significant advances in deep learning techniques, which in turn has substantially advanced other research disciplines, such as natural language processing, image processing, speech recognition, and software engineering. Various deep learning techniques have been successfully employed to facilitate software engineering tasks, including code generation, software refactoring, and fault localization. Many studies have also been presented in top conferences and journals, demonstrating the applications of deep learning techniques in resolving various software engineering tasks. However,although several surveys have provided overall pictures of the application of deep learning techniques in software engineering,they focus more on learning techniques, that is, what kind of deep learning techniques are employed and how deep models are trained or fine-tuned for software engineering tasks. We still lack surveys explaining the advances of subareas in software engineering driven by deep learning techniques, as well as challenges and opportunities in each subarea. To this end, in this study, we present the first task-oriented survey on deep learning-based software engineering. It covers twelve major software engineering subareas significantly impacted by deep learning techniques. Such subareas spread out through the whole lifecycle of software development and maintenance, including requirements engineering, software development, testing, maintenance, and developer collaboration. As we believe that deep learning may provide an opportunity to revolutionize the whole discipline of software engineering, providing one survey covering as many subareas as possible in software engineering can help future research push forward the frontier of deep learning-based software engineering more systematically. For each of the selected subareas,we highlight the major advances achieved by applying deep learning techniques with pointers to the available datasets i

关键词： deep learning software engineering software benchmark software artifact representation survey

来源：评论

学校读者我要写书评

暂无评论

Evaluating the Suitability of Inception Score and Fréchet Inception Distance as Metrics for Quality and Diversity in Image Generation 24

Evaluating the Suitability of Inception Score and Fréchet I...

引用

7th International Conference on Computational Intelligence and Intelligent Systems, CIIS 2024

作者： Chan, Derrick Adrian Sithungu, Siphesihle Philezwini Academy of Computer Science and Software Engineering University of Johannesburg Gauteng Johannesburg South Africa

ISBN: (纸本)9798400717437

Variational Autoencoders (VAEs) have gained popularity as one of the main approaches for generating diverse and high-quality synthetic images. This study examines the suitability of evaluation metrics, specifically Inception Score and Fréchet Inception Distance (FID), for assessing these images. Particularly, the study focuses on the generation of synthetic images based on the MNIST handwritten digits dataset. Through the use of VAE-generated MNIST image samples, the study analyses the abovementioned metrics alongside alternative methods that can be used to assess image quality and diversity. The findings made from the study reveal the strengths and limitations of each metric in evaluating image quality and diversity. This paper underscores the need for tailored metrics to enhance the evaluation of generative models, while specifically using the performance of a VAE as the domain of investigation. © 2024 Copyright held by the owner/author(s).

关键词： Fréchet Inception Distance

来源：评论

学校读者我要写书评

暂无评论

Anomaly Detection in Retinal Imagery Using Variational Autoencoders 24

Anomaly Detection in Retinal Imagery Using Variational Autoe...

引用

7th International Conference on Computational Intelligence and Intelligent Systems, CIIS 2024

作者： Lebaka, Kagiso Blessing Sithungu, Siphesihle Philezwini Academy of Computer Science and Software Engineering University of Johannesburg Gauteng Johannesburg South Africa

ISBN: (纸本)9798400717437

This paper situates itself within the broader medical imaging field, focusing on applying generative modelling techniques for anomaly detection in retinal images. Given the complexity of retinal structures, detecting subtle abnormalities that signal disease requires models that can learn intricate patterns and discern deviations. This study explores the use of a Variational Autoencoder (VAE) to reconstruct healthy retinal images and identify anomalies through reconstruction loss and latent space analysis. With the use of a dataset with an imbalance of healthy and abnormal samples, the model was trained exclusively on healthy images to capture a baseline of normal retinal features. The findings suggest that while the VAE can effectively reconstruct broader structural patterns, it needs to work on finer details, impacting its sensitivity in anomaly detection. These results highlight the challenges inherent in applying generative models to medical diagnostics and emphasise the need for advanced techniques, such as contrastive learning, that may enhance feature extraction. Thus, this paper contributes to the ongoing discourse on the suitability of generative models for medical imaging and supports continued research into more nuanced methods to meet the specific demands of retinal anomaly detection. © 2024 Copyright held by the owner/author(s).

关键词： Ophthalmology

来源：评论

学校读者我要写书评

暂无评论

Automated Pest Control: computer Vision for Wildlife Surveillance 24

Automated Pest Control: Computer Vision for Wildlife Surveil...

引用

7th International Conference on Computational Intelligence and Intelligent Systems, CIIS 2024

作者： Crossling, Cleveland J. Vadapalli, Hima Van Der Haar, Dustin Academy of Computer Science and Software Engineering University of Johannesburg GP Johannesburg South Africa

ISBN: (纸本)9798400717437

The integration of artificial intelligence in agriculture has revolutionized farming practices, enhancing crop yields and resource efficiency. However, existing machine learning systems primarily focus on livestock, overlooking the impact of wildlife on agricultural productivity. This study addresses the gap by introducing two classification pipelines utilizing VGG16 and ResNet50 models for identifying five major classes of wildlife (foxes, rats, pigeons, raccoons, and squirrels) affecting agricultural sectors. A comprehensive dataset, comprising existing datasets and trail-camera footage, is compiled for training and validation. The results show that models trained on clear images struggle to generalize to real-world environments (the best accuracy being 48.19%). Grey-scaled models perform worse with a 2-6% decrease. Training on the training dataset and evaluating the validation set yields better average results for the weaker-performing model. This study contributes to invasive animal control in agriculture, providing a foundation for effective wildlife management with computer vision. © 2024 Copyright held by the owner/author(s). Publication rights licensed to ACM.

关键词： Livestock

来源：评论

学校读者我要写书评

暂无评论

TSN vs I3D vs Pose-C3D: Action Recognition in Basketball using SpaceJam Dataset 24

TSN vs I3D vs Pose-C3D: Action Recognition in Basketball usi...

引用

7th International Conference on Computational Intelligence and Intelligent Systems, CIIS 2024

作者： Chiura, Tafadzwa Blessing Van Der Haar, Dustin Academy of Computer Science and Software Engineering Faculty of Science University of Johannesburg Gauteng Johannesburg South Africa

ISBN: (纸本)9798400717437

Most of the research conducted in action recognition is mainly focused on general human action recognition, and most of the available datasets support studies in general human action recognition. In more specific contexts, such as basketball, datasets that are as comprehensive and publicly available are limited. This study proposes taking three popular and mature methods in the field of action recognition, namely Temporal Segment Networks (TSN), Two-Stream CNN using Inflated 3D-convolutional Neural Networks (I3D) and Pose-C3D, and applying them to the SpaceJam dataset, which is a basketball-specific action dataset. All three experiments used pre-trained ImageNet models and were fine-tuned on the SpaceJam dataset. TSN was the oldest of the methods but obtained the best results of the three experiments, scoring a top-1 and top-5 accuracy of 59% and 96%, respectively. I3D was second best, with a top-1 and top-5 accuracy of 41% and 85%, respectively. Pose-C3D came in third, scoring a top-1 and top-5 accuracy of 15% and 50%, respectively. The results show that the models cannot distinguish significantly between some actions, such as ball in hand, pass and dribble. The study shows that it is feasible for context-specific fine-grain action recognition, but more needs to be done to discriminate against similar actions. © 2024 Copyright held by the owner/author(s).

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Mitigating Ransomware in Government-Managed Institutions: A Global Critical Information Infrastructure Perspective

Mitigating Ransomware in Government-Managed Institutions: A ...

引用

20th International Conference on Cyber Warfare and Security, ICCWS 2025

作者： Mashila, Musiwalo Sithungu, Siphesihle Philezwini Lebea, Khutso Academy of Computer Science and Software Engineering Faculty of Science University of Johannesburg South Africa

ISBN: (纸本)9781917204316

This paper examines the escalating ransomware threats faced by government-managed educational institutions, focusing on their vulnerabilities, case studies, and mitigation strategies. With the adoption of Bring Your Own Device (BYOD) policies, schools increasingly expose their networks to cyber risks, making them attractive targets for cybercriminals. Case studies, including attacks on the Los Angeles Unified School District and the University of California, San Francisco, illustrate the profound impact of ransomware incidents and the diverse responses of institutions. Effective cybersecurity measures are crucial, emphasizing the need for prioritized spending, comprehensive security training, and advanced detection and response strategies. The role of government is also vital, as it develops legislation, guidelines, and funding opportunities to enhance educational cybersecurity. Recommendations include technical measures to secure networks and collaborative educational initiatives to share best practices. Ultimately, this research study underscores the necessity of continuous adaptation and government support in fortifying defences against ransomware threats. By fostering a cybersecurity awareness and resilience culture, educational institutions can better protect sensitive data and ensure the safety of their operations in an increasingly tricky digital landscape. © 2025 Proceedings of the International Conference on Information Warfare and Security (ICCWS). All rights reserved.

关键词： Cyber attacks

来源：评论

学校读者我要写书评

暂无评论

Layout Decomposition via Boolean Satisfiability

引用

IEEE Transactions on computer-Aided Design of Integrated Circuits and Systems 2025年第3期44卷 1112-1125页

作者： Liu, Hongduo Liao, Peiyu Zou, Mengchuan Pang, Bowen Li, Xijun Yuan, Mingxuan Ho, Tsung-Yi Yu, Bei The Chinese University of Hong Kong Department of Computer Science and Engineering Hong Kong Hong Kong Huawei Noah's Ark Lab Hong Kong Hong Kong Institute of Software Chinese Academy of Sciences State Key Laboratory of Computer Science Beijing100190 China

Multiple patterning lithography (MPL) has been introduced in the integrated circuits manufacturing industry to enhance feature density as the technology node advances. A crucial step of MPL is assigning layout features to different masks, namely layout decomposition. Exact algorithms like integer linear programming (ILP) can solve layout decomposition to optimality but lack scalability for dense patterns. Relaxation algorithms (e.g., linear programming and semi-definite programming) and heuristics (e.g., exact cover) are capable of handling large cases at the cost of inferior solution quality. These methods rely on different mathematical solvers and expert-designed heuristics to offer a balance between solution quality and computational efficiency. In this article, we propose a unified layout decomposition framework comprising three algorithms: 1) satisfiability (SAT)-exact;2) SAT-bilevel;and 3) SAT-fast, all leveraging the capabilities of Boolean SAT solvers. The SAT-exact ensures optimality, but with faster convergence than ILP, SAT-bilevel addresses the decomposition as a bilevel optimization problem for rapid near-optimal solutions, and SAT-fast handles very large layouts in an incremental manner. Experimental results demonstrate our framework's superiority over existing state-of-the-art methods in terms of solution quality and runtime. © 2024 IEEE.

关键词： Integer linear programming

来源：评论

学校读者我要写书评

暂无评论

DKGV: A Dynamic Knowledge Graph Visualization Method Based on Force-Directed Layout 19

DKGV: A Dynamic Knowledge Graph Visualization Method Based o...

引用

19th ACM SIGGRAPH International Conference on Virtual-Reality Continuum and its Applications in Industry, VRCAI 2024

作者： Zhou, Hong Huang, Sunjie Zheng, Yushan Liu, Jinyan Chen, Yang Chen, Xiangyuan Li, Jun College of Computer Science and Software Engineering Shenzhen University Shenzhen China Shenzhen Academy of Inspection and Quarantine Shenzhen China Shenzhen University Shenzhen China

ISBN: (纸本)9798400713484

The dynamic knowledge graph is a data structure that adds temporal information to the nodes and edges of a traditional knowledge graph. It describes the changing processes of entities and relationships over time, thereby enriching and updating the expression of knowledge. The dynamic knowledge graph possesses characteristics such as time and labels;it is not only a knowledge graph but also a type of dynamic graph containing heterogeneous information. To address the issue that traditional dynamic graph visualization methods do not account for heterogeneous information in the layout, leading to a relatively random distribution of node labels in the visualization results, this paper proposes DKGV, a dynamic knowledge graph visualization method based on a force-directed layout. DKGV utilizes an improved GCN to generate the initial layout of the dynamic knowledge graph data, performs static force-directed layout iterative processing, and conducts dynamic force-directed layout calculations on the dynamic knowledge graph. Finally, it employs a force-directed layout with boundaries to support three-dimensional temporal display after view transformation. Experimental results show that DKGV can maintain the stability of the node layout during the evolution of the dynamic knowledge graph while making the overall layout more regular to better display the relationships between entities. © 2024 ACM.

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

Universal Construction for Linearizable but Not Strongly Linearizable Concurrent Objects 10th

Universal Construction for Linearizable but Not Strongly L...

引用

10th International Symposium on Dependable software engineering: Theories, Tools and Applications, SETTA 2024

作者： Wang, Chao Wu, Peng Petri, Gustavo Jia, Qiaowen He, Youlin Lv, Yi Liu, Zhiming Centre for Research and Innovation in Software Engineering College of Computer and Information Science Southwest University Chongqing China and State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences Beijing China Arm Research Cambridge United Kingdom China Great Wall Technology Group Co. Ltd. Shenzhen China University of Chinese Academy of Sciences Beijing China

ISBN: (纸本)9789819606016

Strong linearizability is a variant of linearizability and is more suitable for verification. In this paper we investigate the following two problems: (1) for which deterministic sequential specifications there exist linearizable but not strongly linearizable objects;(2) can we capture any violation scheme of strong linearizability for practical objects that are linearizable but not strongly linearizable. To deal with the first problem, we propose two classes of deterministic sequential specifications called inclusively-permutative sequential specifications and exclusively-permutative sequential specifications. The first class of sequential specifications contains sequential specifications of counter, register, max-register, snapshot, set, queue, stack and priority queue. The second class of sequential specifications contains sequential specifications of blocking queue, blocking stack and blocking priority queue. To establish these results, we propose a uniform methodology to construct a wait-free, linearizable but not strongly linearizable object for these two classes of sequential specifications. Our universal construction is based on the classical universal construction, and it additionally permits each operation to change the "take-effect-order" of previous operations. To deal with the second problem, we use bi-branch distinguishable executions to represent violations of strong linearizability. We investigate eight objects that are linearizable but not strongly linearizable, and we show that for each object, it has a pair of bi-branch distinguishable executions such that our universal construction can generate another pair of bi-branch distinguishable executions with the same set of histories. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词：

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：